550 BELL SYSTEM TECHNICAL JOURNAL 



In the absence of hum or other disturbances the build-up of oscillations 

 starts from a randomly fluctuating voltage caused by shot noise. Thus, 

 from turn-on to turn-on some sort of statistical distribution may be expected 

 in the time t taken to reach a given fraction of the final amplitude. In un- 

 published work Dr. C R. Shannon of these laboratories has shown that in 

 terms of «<> , the initial rate of build-up, the standard deviation br and the 

 root mean square deviation (5t')^ are given by 



5t = .38/«o (12.9) 



(572)1/2 ^ ^^^^^ ^2.10) 



Thus the "jitter" in the successive positions of the r-f pulses associated with 

 evenly spaced turn-ons is least when the initial rate of build-up, given by Oo , 

 is greatest. 



Such conditions do not obtain on turn-off, and there is little jitter in the 

 trailing edge of a series of r-f pulses. This is of considerable practical 

 importance. 



XIII. Reflex Oscillator Development at the Bell Telephone 



Laboratories 



For many years research and development directed towards the genera- 

 tion of power at higher and higher frequencies have been conducted at the 

 Bell Telephone Laboratories. An effort has been made to extend the fre- 

 quency range of the conventional grid controlled vacuum tube as w^ell as 

 to ex-plore new principles, such as those embodied in velocity variation 

 oscillators. The need for centimeter range oscillators for radar applications 

 provided an added impetus to this program and even before the United 

 States entry into the war, as well as throughout its duration, these labora- 

 tories, cooperating with government agencies, engaged in a major effort to 

 provide such power sources. The part of this program which dealt with 

 high power sources for transmitter uses has been described elsewhere. This 

 paper deals with low power sources, which are used as beating oscillators in 

 radar receivers. In the following sections some of the requirements on a 

 beating oscillator for a radar receiver will be outlined in order to show^ how 

 the reflex oscillator is particularly well suited for such an application. 



A. The Beating Oscillator Problem 



The need for a beating oscillator in a radar system arises from the neces- 

 sity of amplifying the very weak signals reflected from the targets. Imme- 

 diate rectification of these signals would entail a very large degradation in 

 signal to noise ratio, although providing great simplicity of operation. It 

 would also lead to a lack of selectivity. Amplification of the signals at the 



" See Appendix 10. 



